Freitag, 03. Juni 2022, 15:00 - 16:00 iCal

When to Stop Consulting

 

Hörsaal 1, Erdgeschoß
Oskar-Morgenstern-Platz 1, 1090 Wien

Antrittsvorlesung, Public Lecture


Abstract

 

Consider a decision-maker who must make a binary (Yes-No) decision and wants to maximize the probability of deciding correctly. To gain information, the decision-maker can consult independent experts. But how many? When should consultation stop?

We model the decision-maker’s current state of knowledge using parameters representing the amount of evidence (or advice) in favor of each choice; after each consultation, one parameter value changes. The decision-maker must specifically decide whether to make the decision immediately, based on the advice already received, or to consult once more. We use two approaches to limiting the number of consultations, either a ceiling or a cost, and propose a simple model of the distribution of opinion among the experts.

We find, surprisingly, that it can be optimal not to consult, even when consultation is free – but, less surprising, that consultation is uniquely optimal when the evidence is approximately balanced. We obtain complete optimal strategies when consultation cost is very small and show how the probability of deciding correctly is related to the maximum number of consultations. We compare our results to the Secretary Problem, another decision problem in which sequential information collection is an issue. We also discuss one instance of consultation by a political decision-maker in a major crisis.

 


Veranstalter

Universität Wien, Fakultät für Wirtschaftswissenschaften, Institut für Business Decisions and Analytics


Kontakt

Elitsa Nestorova
Fakultät für Wirtschaftswissenschaften
Institut für Business Decision and Analytics
+43-1-4277-38172
bda@univie.ac.at